Urban stormwater modelling plays a critical role in assessing interventions for flood risk and water quality management in response to ageing infrastructure and future uncertainties. However, modelling workflows in practice remain highly manual, and key steps in model configuration, execution, and interpretation often depend on specialised knowledge, leading to inefficiencies. Therefore, this study proposes SWMM-Agentic, a tool-augmented, large language model (LLM)-based single-agent system for urban stormwater modelling, simulation, and scenario analysis. Built on the Storm Water Management Model (SWMM), SWMM-Agentic uses one orchestration model to interpret natural-language instructions and sequentially invoke documented functions for traceable post-configuration workflows. Evaluation on the Astlingen benchmark included capability demonstrations and a 60-task suite comprising 20 static, 20 dynamic, and 20 scenario-based tasks, executed once with each of three LLMs to produce 180 model-task runs. DeepSeek-V3.2-Exp successfully completed 59/60 tasks (98.3%), Qwen3-236B completed 58/60 (96.7%), and Qwen3-14B completed 45/60 (75.0%). Across 180 runs, 89 of 100 failed tool calls were followed by a successful corrective call within three attempts. SWMM-Agentic also reproduced network characteristics, compared alternative control strategies, and conducted a human-framed rain-garden experiment that showed decreasing combined sewer overflow discharge with diminishing marginal benefits at higher coverage. These results demonstrate that SWMM-Agentic can reliably operate existing SWMM models through natural language within the evaluated benchmark and tool scope, supporting accurate and reproducible stormwater simulation and analysis, and laying the groundwork for natural-language-driven platforms for integrated planning and hypothesis-driven research.
Jian Wang, Chenyue Sun, Dragan A. Savić et al.· Journal of Environmental Man...· 0 citations
Biofilm development in drinking water distribution systems (DWDS) affects water quality, hydraulic performance, and microbial risk, yet its spatial distribution and structural properties remain poorly characterized. Existing assessment methods rely on microbiological or bulk water indicators that are difficult to interpret at the system scale and do not directly reflect biomass accumulation on pipe walls. This study presents the first model to predict the mean biofilm thickness in drinking water pipes using routinely measurable operational variables under controlled laboratory conditions. Biofilms were grown in a controlled pipe facility over ten months, and thickness was quantified using a hydraulic residence time method. A random forest model using seven variables describing hydraulic, thermal, and limited chemical conditions achieved high prediction accuracy (R2 = 0.91 on unseen data) and identified flow rate, water temperature, and environmental stability as dominant factors. Feature importance and SHAP analyses highlighted the influence of conditioning shear stress and recovery time, while meta-analysis showed how operational conditions govern the formation of a stable biofilm base and a more easily removable outer layer. By linking operational conditions to biofilm thickness, this work provides a foundation for assessing and managing biofilm accumulation in drinking water systems subject to further validation.
Konstantinos Glynis, M. Blokker, Z. Kapelan et al.· ACS ES&T Water· 0 citations
Urban water systems are increasingly challenged by climate extremes, aging infrastructure, and rising flood risks. Conventional water management practices remain fragmented across data, operations, and assets, limiting coordinated decision-making and scalable engineering deployment. Digital twins (DT) show great promise to overcome this fragmentation for resilient and efficient management. This review proposes an engineering-practice-oriented framework of digital twins for smart water management (DTSW). Utility demands are first structured through a scenario-oriented decomposition into points of interest (POIs), thereby linking practical engineering problems to digital variables. The review further summarizes a probabilistic graphical model-based scheme as the algorithmic backbone for POI implementation, and examines the key enabling technologies across organized data foundations, models, and real-time control. Particular attention is given to AI-empowered DTSW techniques, including soft sensing and data cleansing, hybrid modeling, and uncertainty-aware model deployment. Future development is discussed from the perspectives of proactive optimization, human-digital collaboration, and scalable engineering deployment. This review thus provides a structured framework for guiding the practical design and deployment of DT in urban water systems, facilitating coordinated, scalable and resilient water management.
Haozheng Wang, Jinkuo Li, Xuhui Dang et al.· Water Research· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
